Secondary data analyses on two analogous data sets assessed the abilit
y of crime-scene variables observed in serial rapes to predict classif
ication in the Massachusetts Treatment Center, Rapist Typology Version
3 (MTC:R3). Data sets from the FBI containing extensive coding of cri
me-scene information but minimal offender data were analyzed to select
optimal predictors by examining the frequencies and both within-crime
and across-crime consistencies of crime-scene indicators. Analogous p
redictors were identified in the MTC database, which contains extensiv
e offender data but minimal crime-scene data, and these indicators wer
e employed to predict rapist type. Promising predictive results emerge
d in the domains of adult antisocial and expressive aggression. In add
ition, the domains of sadism, offense planning, and relation with vict
im showed high internal consistency and good to high cross-crime consi
stency, suggesting that predictive scales are possible for these domai
ns.